Graph-Based Pattern Identification from Architecture Change Logs

Service-based architectures have become commonplace, creating the need to address their systematic maintenance and evolution. We investigate architecture change representation, primarily focusing on the identification of change patterns that support the potential reuse of common changes in architecture-centric evolution for service software. We propose to exploit architecture change logs capturing traces of sequential changes to identify patterns of change that occur over time. The changes in the log are formalised as a typed attributed graph that allows us to apply frequent sub-graph mining approaches to identify potentially reusable, usage-determined change patterns. We propose to foster the reuse of routine evolution tasks to allow an architect to follow a systematic, reuse-centered approach to architectural change execution.